I'm looking for any recommended approaches to be able to digitally process a room full people, like an auditorium or a movie theater and output the following information:

  • How many people are in the room
  • Locate unoccupied and available seating
  • Display the available seating to ushers, or people attending late, on an iPad or TV outside the auditorium. Similar to viewing airplane seat occupancy online.

Some sensors I'm considering

  • LiDar Scanning
  • Many ceiling (IR) cameras looking straight down (or perpendicular to the seating if the seating is inclined)

Some Challenges: Be able to take accurate enough measurements when:

  • The lights are on
  • The lights are off or in low light
  • People are standing
  • People are sitting
  • Seating is a long bench, not a chair


  • Individual sensors for each seat isn't ideal, as we need to count people when they are standing as well.

Thoughts on LiDar:

I like the thought of scanning the room with LiDar to develop a 3D Map from which I can determine the number of prominent heads to count, and distance/location of each head to identify a seat that is taken or open.

Are there any affordable (under $10,000) LiDar scanners that make this easy?

Thoughts on Top Down Infrared (IR) Cameras :

The auditorium seats 700 people and has a balcony. If the cameras are positioned in a top-down perspective, we would need to install about 22 cameras. It would be nice to have less of an installation overhead. Also, we'd like to respect the audience and not look down shirts.

Thoughts on 1 or 2 Infrared (IR) Cameras Viewing the Crowd from the stage:

This could work for counting faces (using OpenCV or other Computer Vision library) but would make it difficult to tell what seats are open and available, especially when people are standing.

Images of the Auditorium:

enter image description here enter image description here enter image description here

  • $\begingroup$ Are you going to ask the people if they don't mind being filmed? $\endgroup$
    – FooTheBar
    Sep 16 '17 at 14:05
  • $\begingroup$ Why using lidar or camera when you could just put a sensor on each seat to know if they are down or not ? $\endgroup$
    – Malcolm
    Sep 17 '17 at 8:48
  • $\begingroup$ @Malcom Hi Malcolm, because we'll need to count and know the location of people when they are standing as well. $\endgroup$
    – aero
    Sep 18 '17 at 14:41
  • $\begingroup$ Then, it's too different processes: count the number of people, and count the number for seats used. Isn't it ? You could separate the two. $\endgroup$
    – Malcolm
    Sep 20 '17 at 10:43
  • 1
    $\begingroup$ Although this question is too old to migrate now, it would have been more appropriate over on Signal Processing. $\endgroup$
    – Mark Booth
    Apr 11 '18 at 14:41

A Near-IR or IR camera which can rotate and tilt, that is installed over the stage would probably be an acceptable and fairly well performing solution.

Installation would be simple: screw on, and take the cable down (hidden somehow), to a computer system for processing.

This would take several pictures/or continuous video and an image processing software can determine how many people are seated.

One particular point is that, the seats are in fact long and non-discrete sofas. So, the software should assume a capacity for each of the seats.

If this is a one time installation, you could sort things out on the way. However, if this is for a product to be installed at several stations, more generic systems would be preferred.

  • $\begingroup$ Thanks for your answer Gürkan Çetin. I imagine it will be difficult to tell which row people are sitting in at the back of the auditorium. Is there a way for an IR camera to accurately (within a 50 cm) measure distances? $\endgroup$
    – aero
    Sep 16 '17 at 19:06

Consider normal COTS video feed under good lighting conditions and using OpenCV to count people. You would have to make general assumptions about what qualities make an image of a person. Find how well it works. Then refine your assumptions. You may want to eventually try training OpenCV. There exist OpenCV examples files for face recognition, cat recognition, eye recognition, ect. However, in your case, you need body recognition from above.

Likely, over time, you will develop a complex procedure to identify people, stop counting people in walk ways and tell the difference between a person and a purse.

The initial attempt may be to record an empty room, subtract that image from an image of a full room and detect / count the areas which are different.

  • $\begingroup$ Thanks for your answer st2000. Some challenges I have are times when the auditorium will have no/low light. Also, I also need to determin if there are open seats (location of people) not just counting people. I do like the idea of training OpenCV to recognize people over purses. $\endgroup$
    – aero
    Sep 16 '17 at 18:50
  • $\begingroup$ Think of OpenCV as a library of image tricks geared to understand and locate items in a picture. The more your practice, the better you will get at using OpenCV. If the lighting conditions change, I image the effectiveness will be crippled. However, as I said, use a base picture of an empty bench but this time with low light when looking for people in low light conditions. BTW, benches are not individual seats. You will have to form your question better as to what constitutes a vacancy both for this question and for your self. $\endgroup$
    – st2000
    Sep 16 '17 at 19:37
  • $\begingroup$ No light? That is nearly impossible with COTS cameras. Likely it will demand such an investment in multiple IR cameras (and possibly IR constructive light projectors) that only the most exotic venues would even consider it. Or, you could remove the IR filter from a COTS camera. But beware that IR cameras can make some materials somewhat transparent. $\endgroup$
    – st2000
    Sep 16 '17 at 19:45

Not the answer you're looking for? Browse other questions tagged or ask your own question.